No-regret Algorithms for Fair Resource Allocation

Authors: Abhishek Sinha, Ativ Joshi, Rajarshi Bhattacharjee, Cameron Musco, Mohammad Hajiesmaili

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we report experimental results using the fair caching problem as a case study. Additional experiments on fair scheduling are provided in Appendix D.2. We compare the performance of our algorithms on two datasets against several baselines showing the effectiveness of our algorithm.
Researcher Affiliation Academia Abhishek Sinha, Ativ Joshi School of Technology and Computer Science Tata Institute of Fundamental Research Mumbai 400005, India abhishek.sinha@tifr.res.in ativ@cmi.ac.in Rajarshi Bhattacharjee, Cameron Musco, Mohammad Hajiesmaili Manning College of Information and Computer Sciences University of Massachusetts Amherst {rbhattacharj, cmusco, hajiesmaili}@cs.umass.edu
Pseudocode Yes Algorithm 1 The Online Fair Allocation (OFA) Policy
Open Source Code Yes The code is available at: https://github.com/Ativ Joshi/nofra
Open Datasets Yes We perform simulations on both synthetically generated data and on CDN traces from Berger [2018].
Dataset Splits No Guaranteeing fairness becomes even more challenging in the online learning setup as there is no distinction between the training and test data, and no assumption is made on the input data sequence.
Hardware Specification No The paper does not provide specific hardware details used for running experiments.
Software Dependencies No The paper does not list specific software dependencies with version numbers.
Experiment Setup Yes For our synthetic dataset, we take time horizon T = 1000 rounds, number of users m = 5, cache size C = 7, and library size N =30. At each time step: (1) users 1 and 2 request a file from file numbers 0-29 uniformly at random; (2) user 3 repeatedly requests file numbers 0-3; (3) user 4 repeatedly requests file numbers 4-18; (4) user 5 repeatedly requests file numbers 19-27.